Pinned Tweet
Acceptance: “I’m here for reality. Whatever’s real, bring it. I’ll deal with it.”
Acceptance in the buddhist or stoic sense is *not* acceptance of a particular narrative.
Acceptance is deference to the world. Dynamic responsiveness to dynamic reality is boss-level Acceptance.
This was not written by a bot 🤪
The term alignment seems vague. It is undefinable. A better term is maybe accountability
We can design for accountability, but not alignment.
By keeping systems transparent and simple, even contributors that are more intelligent can be held accountable by keeping the plans and processes visible.
Where we lose visibilty, accountability disappears.
The Nash Equilibrium aficionados are living this weird “max for me” systematized selfishness, leaving discretionary effort on the table. Everybody loses the extra benefits they could get by cooperating, making and keeping agreements and alliance, and building trust
Nash straight-up contradicted von Neumann, maybe thinking we wouldn’t / couldn’t notice.
Nash: trust doesn’t exist, everyone serves themself.
Von Neumann: obviously we have to trust each other. And we have to build trust snd keep it.
“This is the core of the Nash equilibrium: rational action is whatever maximizes your own expected payoff, assuming every other player does the exact same thing independently. There is no room for trust, promises, or joint strategy unless it happens to be self-enforcing in equilibrium.”
Nash is a fraud 😆 omg. Gah
Not everyone thinks like this
And NONE of them will wash your laundry or make food for you.
The taking children seriously guys are aiming at less coercion. But, if you think of family as community, what would be a better community design: “let’s be happy together and hold each other account for that” vs “let’s all do whatever the heck we want”?
Clearly, the being-happy setting is more fun than the do-whatever-we-want situation
Note well: the apricots in the store ($14 for 3lbs) are not the same as the $1k/ day apricots I used to consume as a kid hanging out in our 3 apricot trees. We got those for “free”, if we would brave the vigilance of two cows who also wanted to eat the free apricots. We would always brave it. Barefoot and brazen.
Excellent summer days, dangling from the trees, harvesting unlimited tree-ripened fruit from its twig, and outwitting the territorial pasture-raised cows.
A store-bought apricot cannot compare.
2 areas are exquisitely susceptible to AI disruption: hardware built to be remotely operated (drones). Purely theoretical research (math, programming, algorithm execution)
Everything else is: a) hard to do, b) hard to automate, c) resistant to disruption, d) skill dependent.
Addendum: any real world building requires investigation of the specific site, which takes time for a human and a crazy amount of engineering for an automated investigator to parse. If you move the investigation out of bits and into reality, human intelligence wins for the medium term because the actuators (arms and hands) already work with the processor (brain) without any additional R&D.